Page 63 - Psychological Management of Individual Performance

42 ability and non-ability predictors of job performance
As such, meta-analytic findings on general cognitive ability–performance relations
must be interpreted in terms of a mismatch between the predictor (a test of “maximal”
performance) and the criterion (a measure of “typical” performance). Predictive va-
lidities in these studies thus represent estimated true-score relations between what an
individual is capable of under maximal effort conditions and what an individual does
under typical job conditions, and so may under- or overestimate the influence of general
mental ability for job performance.
In a step that moves toward addressing this problem, Ackerman (1994) suggested that
the psychometric premise that intelligence is best associated with maximal performance
is incomplete. He argued that measures of intelligence often contain both typical and
maximal properties, and that such measures may be integrated with non-ability mea-
sures using an intelligence perspective that distinguishes between maximal and typical
measures of ability in terms of intelligence as process (maximal) and intelligence as
knowledge (typical). Though such a perspective has yet to gain widespread attention
in the I/O research literature, the trend toward integrated assessment of person factors
suggests that this issue will receive increasing attention in the future. In the I/O domain,
personality measures appear to provide significant incremental validity when the cri-
terion of interest involves job behaviors other than technical performance (e.g., inter-
personal skills, employee dependability). As such, it appears reasonable to expect that
the integrated use of ability (maximal) and personality (typical) measures will be most
helpful when the criterion includes consideration of an individual’s typical behaviors in
the broader work context as well as performance of designated tasks.
Meta-analyses of personality–job performance relations avoid the typical–maximal
form of construct mismatch, but are vulnerable to two other forms of construct mis-
match. First, as noted by Wittmann and Suess (1999), a mismatch may occur in terms of
the breadth of the predictor and criterion spaces. In personality–performance research,
for example, a broad measure of conscientiousness may be used to predict a narrow mea-
sure of technical job performance. Such forms of mismatch lead to predictive validities
that may under- or overestimate relations among the underlying trait and performance
constructs. Second, as noted by Murtha, Kanfer, and Ackerman (1996), dimensions of
personality may not be consistent across environments. For example, individuals who are
conscientious in home environments may be considerably less so in work environments.
Personality measures designed to assess cross-situational personality traits may provide
poor prediction of such behaviors in a particular environment, such as work.
A second concern pertains to the field’s heavy reliance on the meta-analytic framework
for extending knowledge about person–performance relations. Standard meta-analytic
methods assume linearity between the predictor and criterion. In personality–job perfor-
mance relations, the assumption of linearity may be inappropriate when the underlying
function between the personality dimension and the performance criteria are non-linear
(cf. Eysenck, 1995; Snow, 1989). In the workplace, for example, extroverted tendencies
might cause individuals to spend too much or too little time talking with coworkers, con-
tributing to lower levels of job performance compared to individuals who show moderate
extroversion tendencies.
A related issue in the personality–performance domain pertains to the potential bidi-
rectionality of personality constructs. As Tett et al. (1999) noted, personality (and other
non-ability measures) differ from ability measures in their potential for bidirectionality.